CLINICAL CALORIMETRY

TENTH PAPER

A FORMULA TO ESTIMATE THE APPROXIMATE SURFACE

AREA IF HEIGHT AND WEIGHT BE KNOWN

DELAFIELD Du BOIS, B.S.

BURLINGTON, VT.
AND

EUGENE F. Du BOIS, M.D.

NEW YORK

Since the publication of Paper 5 of this series the so-called "Linear

Formula" has been used in the study of a large number of individuals.
Practically all of the subjects of respiration experiments in the Sage
calorimeter have been measured in this way, and in addition Means1
of Boston has used it as a factor in determining his normal base line of
metabolism and the extent of the pathological variations. Means has
found that the range of normal variation from the average is smaller
and that the apparent depression of metabolism in obesity is much less
marked when the linear formula, instead of Meeh's formula, is used
to determine surface area.
The accuracy of the linear formula has been shown in Paper 9 of
this series. In order to correct the slight error in the factor for the
arms, and also in order to clear up a few points in the measurements
which may cause confusion, it seems best to repeat the formula and
show the bony landmarks by diagram (Fig. 1). Some difficulty has
been experienced in locating the superior border of the great trochanter
in fat subjects. This landmark is the starting point of the measure¬
ment "O" which represents the length of the thighs. If we employ
another factor we can use the new measurement "W," the distance
from lower border of the patella to the upper border of the pubes, a
point already located in the measurement "L." In taking this measure¬
ment, however, one must be careful to have the legs straight and the
knees, heels and great toes touching. It is better to take all measure¬
ments from a footboard with the subject lying down,2 determining dis¬
tance from soles of feet to lower border of patella, to upper border of

Submitted for publication Feb. 4, 1916.

From the Russell Sage Institute of Pathology, in Affiliation with the Second
Medical Division of Bellevue Hospital.
1. Means, J. H.: Studies of the Basal Metabolism in Obesity and Pituitary
Disease, Jour. Med. Research, 1915, xxxii, 121; Basal Metabolism and Body
Surface, Jour. Biol. Chem., 1915, xxi, 263.
2. This is especially important with obsese patients.

Downloaded From: http://archinte.jamanetwork.com/ by a University of Arizona Health Sciences Library User on 08/31/2013
 pubes, to suprasternal notch and to top of the head.3 In Table 1
a comparison is made of the old and new formulas for determining
the surface of the thighs. It is seen that the average error is the same.
In the literature of the work on respiratory metabolism it has been
customary to give only the age, weight and height. If, therefore, we

Fig. 1.—Measurements used in "Linear Formula."

are to recalculate previous work in an effort to get more accurate results
than are furnished by Meeh's formula we must content ourselves with
calculations based on height and weight. A formula such as Meeh's,
based on weight alone, can easily give an error of 15 to 20 per cent.,
3. Dr. F. G. Benedict of Boston has called our attention to the fact that this
determines the length rather than the height. We have found that as a rule
the length is 1 or 2 centimeters greater than the height, but we must remem-
ber that height varies 1 to 3 centimeters during the day.

Downloaded From: http://archinte.jamanetwork.com/ by a University of Arizona Health Sciences Library User on 08/31/2013
 ScJ313WliN30-lH3üH

Downloaded From: http://archinte.jamanetwork.com/ by a University of Arizona Health Sciences Library User on 08/31/2013
 but this error is greatly reduced by taking the height into consideration.
With people of very unusual body shape there does not seem to be
any accurate method simpler than the linear formula with its nineteen
measurements. The reason why a consideration of the height does not
entirely correct the calculations based on weight becomes apparent
when we consider the circumference of the body at various levels. For
instance, in the case of R. H. H. the average circumference of the legs
was 30.0 cm. and of the thighs 43.9 cm. An increase of 10 cm. in the

length below the knees would mean an increase of 600 sq. cm. in surface
area, but if the length of the thighs were increased 10 cm. it would
mean a gain of 878 sq. cm. Variations in the arms would not affect
the height at all.

TABLE 1.—Comparison of Old and New Formulas for Determining

Surface Areas of Thighs

Thigh Measurement* Surface Surface Surface

Thighs Cale. Error, Cale. Error,
Name "O" "W" asMeas., 0(P+Q) W(P+Q) %
Cm. Om. Sq. Cm. 0.508 0.552

Benny 1. 26.4 22.4 1284 1294 +1 1195 7

-

Morris S. 41.7 38.8 3022 3207 +0 3251 + 7

R.H.H. 47.0 46.0 3712 3512 —5 3745 + 1
E. F.D. B. ... 46.3 46.3 3820 3655 —4 3981 + 4
Mrs. McK. 40.0 32.2 3500 3594 -3 3152 —11

Anna M. 16.0 14.4 47S 4SS +2 479 + 0

GeraldS. 44.7 44.9 3002 2677 —11 2927 3

—

Emma W. 45.7 41.7 3324 3448 +4 3425 + 3

E.H. S. 50.3 40.3 3175 3457 +9 3464 + 9

Average.
*
Old measurement "O," superior border of great trochanter to lower border of patella.
New measurement "W," superior border of pubes to lower border of patella.

A formula to express surface area must naturally be a bi-dimen-

sional formula, as surface involves two dimensions. If we assume
that weight is proportional to volume, it is obvious that three dimen¬
sions are involved in any expression for weight. Height is, of course,
a single dimension. If we attempt to construct a formula for surface
area (A) based on weight (W) and height (H), it is obvious that a

simple formula such asA WXH~XC (C being a constant depend¬

=

ing on the units used and the subject to which the formula is to apply)
is not logical. In this formula one side, A, is bidimensional and the
other side, W H X C, involves four dimensions, three from W
and one from H. If W is tridimensional, it is obvious that the cube

Downloaded From: http://archinte.jamanetwork.com/ by a University of Arizona Health Sciences Library User on 08/31/2013
 root of W (=%/W or W1/3) is undimensional and a formula
A = W1/3 H X C is logical in that it is bidimensional on both sides.
Another bidimensional expression involving W and H would be the
square root of W X H (V W H or W1/2 X W2) because W X H,
being four-dimensional, is reduced to a bidimensional expression on
taking the square root. A formula based on this method of reduction
would be A = W1/2 X H1'2 X C.
TABLE 2.—Measurements and Constants for Linear Formula (Measure¬
ments Taken with Subject Lying on a Flat Surface)

Head: AB 0.308.
A—Around vertex and point of chin.
—Coronal circumference around occiput and forehead, just above eyebrows.
Arms: F(G + H + 1) 0.611.*
F—Tip of acromial process to lower border of radius, measured with fore¬
arm extended.
G—Circumference at level of upper border of axilla.
H—Largest circumference of forearm (just below elbow).
I—Smallest circumference of forearm (just above head of ulna).
Hands : JK 2.22.
J—Lower posterior border of radiustip of second finger.
to
—Circumference of open hand meta-carpo-phalangeal joints.
at the
Trunk (Including neck and external genitals in the male, breasts in female) :
L(M + N) 0.703.
L—Suprasternal notch to upper border of pubes.
M—Circumference of abdomen at level of umbilicus.
—Circumference of thorax at level of nipples in the male and just above
breasts in the female.
Thighs: 0(P + Q) 0.508.
O—Superior border of great trochanter to the lower border of the patella.
—Circumference of thigh just below the level of perineum.
Q—Circumference of hips and buttocks at the level of the great trochanters.
Or:—Thighs: W(P + Q) 0.SS2.
W—Upper border of pubes to lower border patella (measured with legs
straight and feet pointed anteroposteriorly).
—As above.
Q—As above.
Legs: RS. 1.40.
R—From sole of foot to lower border of patella.
S—Circumference at level of lower border of patella.
Feet: T(U + V) 1.04.
—Length of foot including great toe.
U—Circumference of foot at base of little toe.
V—Smallest circumference of ankle (just above malleoli).
*
Factor 0.558 if F is measured over olecranon with forearm flexed.
Note.—-The constants for arms, thighs, etc., when multiplied by the measure¬
ments of one side give the surface area for both sides. To find total surface
area add the seven parts.

Downloaded From: http://archinte.jamanetwork.com/ by a University of Arizona Health Sciences Library User on 08/31/2013
 TABLE 3.—Comparison of Various Formulas

Measured Area Area Area

or = C = C
Deter¬
Ht.^Wt. -

VHt.XWt. -
l
mined by
Height Linear Wt^XHt.1
Name Weight, or For¬
Kg. Length, mula, Varia¬ Varia¬
Cm. Area, Factor tion Factor tion Factor
Sq. C from C from C
Cm. 25.6, 167.2,
% %

Measured by
Molds:

Benny L. .. 24.2 110.3 S473 20.7 +4.3 164.0 -2.0 72.30

Morris S. .. 64.0 164.3 16720 25.5 —0.5 163.0 —2.5 70.65
R. H. H. ... 64.1 178.0 18375 25.8 +0.8 171.5 +2.6 73.22
E F. D. . 74.1 179.2 19000 25.3 —1.2 164.7 —1.5 70.85
Mrs. McK. 93.0 149.7 18592 27.2 +6.2 157.5 —5.8 71.70
.

Gerald S... 45.2 171.8 14901 24.4 —4.7 169.2 +1.2 70.36
Fab. S. 32.7 141.5 (11869)' 25.5 (-0.5)' 174.2 (+4.0)' 74.37
Anna M. ... 6.27 73.2 30!» 27.4 +7.1 172.0 +3.0 75.54
Emma W. . 57.6 164.8 16451 25.8 +0.8 169.0 +1.1 72.56
R. H. S. 63.0 184.2 17981 24.5 —4.3 167.0 0.0 70.58
Average- +3.3 +2.2
Measured by
Linear
Formula
Edw. . 62.3 174.0 17270 25.0 —2.3 165.2 —1.2 70.75
John . 66.4 176.0 17610 24.9 —2.7 164.2 —1.7 70.83
Alb. S. 66.4 162.2 16720 25.5 —0.5 161.0 -3.7 70.20
Wm. S. 44.6 179.0 15450 23.7 —7.4 172.8 +3.3 71.53
A.F.C. 69.6 179.4 17960 24.4 —4.7 160.5 —4.0 68.71
Wm. A. 63.4 180.0 17940 24.9 —2.7 168.2 +0.6 71.22
Mart. C. 44.0 166.8 14370 24.5 —4.3 167.5 + .1 70.44
Jos. U. 40.1 179.0 14520 23.7 —7.4 171.2 +2.3 70.36
Wm. Shee_ 63.8 171.0 16070 23.5 —8.2 154.8 —7.5 66.06
Arthur V. ... 58.3 155.0 15560 25.8 +0.8 163.3 —2.4 73.02
Armon W. 60.8 161.0 15500 28.6 —7.8 156.6 —6.4 67.96
...

Annie . 26.3 137.0 10460 25.0 0.0 174.0 +4.1 73.60

Fred D.. 49.3 157.0 13870 24.1 —5.8 157.2 —6.0 67.69
Fred D. 40.0 157.0 12960 24.2 —5.5 163.5 —2.3 69.12
Edw. . 50.4 168.0 14830 23.4 —8.6 161.0 —3.7 68.27
J. McE. 41.8 166.0 13260 23.1 —9.7 159.0 —4.8 66.67

Bart D. 43.5 156.0 13850 25.3 —1.2 168.1 +0.5 71.62

Burr Ph. 70.7 169.0 17390 24.9 —2.7 159.0 —4.8 69.01

J. D. D. B.... 34.5 152.8 12240 24.7 —3.5 168.6 +0.7 70.90

Downloaded From: http://archinte.jamanetwork.com/ by a University of Arizona Health Sciences Library User on 08/31/2013
 TABLE 3.—(Continued)
Measured Area Area Area
or = C = C = C
Deter¬ Bt.v'Wt. -

VHt.xWt. -
-

Height mined by Wt. XHt.

Name Weight, or Linear
Kg Length, For¬
Cm. mula, Vaca¬ Varia¬ Varia¬
Area, Factor tion Factor tion Factor tion
Sq. C from C from C from
Cm. 25.6, 167.2, 71.84,
% %

Ray M. ... 30.4 140.5 10840 24.8 —3.1 165.8 -0.8 70.42 —1.9

Harry B. 36.5 146.0 12320 25.4 —0.8 168.6 +0.8 72.01 +0.3
12240 71.30 —0.8
.

Harry K. . 35.9 148.2 25.0 —2.3 167.8 +0.2

Arthur A. 30.6 146.6 11260 24.0 —3.9 167.0 —0.1 70.71 —1.6

Leslie B. .. 2S.5 140.8 10500 24.5 ^.3 166.5 —0.5 70.00 —2.6

Peter N. .. 03.4 187.7 17950 24.0 -6.2 164.2 —1.8 69.09 —3.8

Max W. .. 73.2 173.7 18540 25.5 —0.5 164.5 —1.7 71.07 —1.1
Dan O'C. 60.0 167.0 16570 25.4 -0.8 165.7 —1.0 71.12 —1.0
.

Jack O'C. 31.4 162.8 12040 23.5 —8.2 168.2 +0.6 69.35 —3.5
A. F. 52.0 159.0 14870 25.0 —2.3 162.5 —2.9 69.95 —2.6
G.L. 79.2 175.5 20300 26.9 —5.1 172.2 +3.0 74.70 +4.0
F. C. G. .. 50.5 173.9 16440 24.7 —3.5 167.4 0.0 70.29 —2.1
L. M. 59.5 170.6 16340 24.6 —3.9 162.1 —3.0 69.30 —3.5
F. G. . .. S7.1 182.8 20760 25.8 —0.8 164.5 —1.1 71.22 —0.9

*
Measured by adhesive plaster method which gives results about 3.3 per cent, too high.
The plus variations would be reduced by this amount.

Comparing the two A formulas

W1/3 C and =

A = W1/2 H1/2 X C, it will be that they differ in the relative

seen

importance given to W and H. In the former W has less importance

and more importance than in the latter. Meeh's formula
A = W2/3 C, failed because was neglected entirely. Adding
to the formula makes it more nearly applicable to subjects of the same
general shape but differing somewhat in relative dimensions, and the
best formula involving only W and will be the one which gives a
certain best relative importance to W and H.
Both of the above formulas were carefully investigated by applying
them to the nine subjects that had been measured in the laboratory.
For these subjects W, H4 and A are known. In testing a formula
the procedure was to solve for C (the only unknown) for each
of the ten cases and then to assume the correct constant for the
formula to be the average value of the C's so found. The merit of the
4. With about half the subjects this was determined standing. No attempt
has been made to correct for the difference between height and length. (See
Footnote 3, p. 864.) The largest difference we have found would cause a change
of about 1.5 per cent, in the surface area and reading. This is within the limit
of accuracy claimed for the Height-Weight Chart.

Downloaded From: http://archinte.jamanetwork.com/ by a University of Arizona Health Sciences Library User on 08/31/2013
 formula was then judged by the percentage variation of the factors
C, as found for the individual cases, from the constant chosen. This
percentage variation would also be the percentage error in area in
the individual cases if the formula were applied using the chosen
constant.
The formulas with H1 and H1/2 both gave rather good results, but
itwas noticed in a number of cases that the percentage error for the
same subject differed in sign for the two formulas. This would indi¬
cate that some formula would be better than either of these two if H
were raised to some power between y2 and 1.
The formula A = W1/3 H X C can also be written
A =W1/3 X H171 X C, bringing it into the same form as
A =W1/2 X H1/2 X C and the general form of this formula can be
written A W1/a X H1^ X C. In order that the expression
=

W1/a X H1/b C may remain bidimensional it is only necessary that

3/a X 1/b = 2, as it does in the two cases considered. For an inter¬
mediate equation it is obvious that (b) must be greater than 1 but less
than 2. A value of b = 1.25 would give a 2.5 and the formula
=

would be A W1'2·5 H1/1·25 X C. This formula when tested gave

=

very much better results than either of the others, but to find the best
values of "a" and "b" it was necessary to explore formulas having a
number of other combinations of "a" and "b" and then to interpolate
graphically.
The best values of "a" and "b" were found to be
a 2.35 and b = 1.38 giving the formula the final form of
=

A =1/2.35 Hi/i.38 c or A W0·425 = 0·725 X 71.84. This for¬

mula can be solved by logarithms as follows :
Log. A Log. W X 0.425 + Log.
=
0.725 4-1.8564.
1.8564 is a constant equal to Log. C.
In order to make this somewhat complicated formula easy of appli¬
cation a chart has been constructed (Fig. 2). By means of this it
is possible to find the approximate surface area at a glance. The
ordinates represent the height in centimeters, the abscissae the weight
in kilograms. The point of intersection of these lines is found for any
given subject and the surface area in square meters read off on the
curved lines by interpolation. The second decimal place, which is
never accurate, is estimated by the distance of the point from the
nearest curved line. For instance, if the man were 150 centimeters
tall and weighed 60 kilograms, the approximate surface area would be
1.55 square meters.
The large plus error in the constant employed by Meeh5 has been
established by the previously quoted works of Bouchard, Lissauer,
Sicheff, Lassabliere and ourselves. According to our calculations the
5. Meeh: Ztschr. f. Biol., 1879. xv, 425.

Downloaded From: http://archinte.jamanetwork.com/ by a University of Arizona Health Sciences Library User on 08/31/2013
 average error in Meeh's constant is about 15 per cent. Instead of
a uniform figure of 12.312, the "constant" should average about 10.5,
varying between 12.3 for the greatly emaciated and 9.0 for the very
stout. We must remember that figures for the calories per square
meters of body surface will average 15 per cent, smaller when Meeh's
formula is used then when the linear formula or the new Height
Weight Formula" is employed.
SUMMARY AND CONCLUSIONS

The method of calculating the surface area from the so-called

"Linear Formula" is given with a slight correction in the factor for
the arms and an alternative measurement for the thighs. A simpler
"Height-Weight Formula" has been devised to estimate the surface
of subjects if only their height and weight be known. This is expressed
in the terms A = W0·425 0·725 C. A being the surface area in
square centimeters, the height in centimeters, W the weight in kilo¬
grams and C the constant, 71.84. A chart has been plotted from this
formula so that the approximate surface area may be determined
at a glance.
We may estimate the errors in the various formulas as follows :
"Linear Formula" and "Height-Weight Formula," maximum ± 5
per cent., average ± 1.5 per cent., Meeh's Formula, maximum -\- 30
per cent., average +15 per cent. In general the maximum figures
apply only to those of unusual shape, while with those of average
body form the average error will seldom be exceeded.
Converse Hall—477 First Avenue.

Downloaded From: http://archinte.jamanetwork.com/ by a University of Arizona Health Sciences Library User on 08/31/2013